Multi-Model AI & BYOK Architecture - Quixa

Quixa adopts a Bring Your Own Key (BYOK) enterprise AI architecture. All AI inference calls are proxied securely through the Quixa backend with tenant-level credential isolation—your API keys are encrypted at rest and never exposed to client browsers.

1. Supported Model Catalog

Quixa dynamically filters and serves top-tier LLM providers based on tenant feature flags:

Provider	Supported Models	Recommended Use Case
Google Gemini	`gemini-3.5-flash-lite`, `gemini-3.5-flash`, `gemini-2.5-flash`	Default Recommended: High-speed, ultra-low-latency flow generation and test synthesis.
Anthropic Claude	`claude-3-7-sonnet`, `claude-3-5-sonnet`	Complex multi-branch reasoning and intricate JavaScript transformations.
OpenAI	`gpt-4o`, `gpt-4o-mini`	Enterprise-standard structured outputs and scenario formulation.
Ollama (Self-Hosted)	Local Llama 3 / DeepSeek / Mistral	On-premise air-gapped environments with zero external network egress.

2. Managing AI Credentials (BYOK)

Platform and Tenant Admins configure AI credentials in the Backoffice Control Panel:

Navigate to Backoffice -> Telemetry & AI (or Workspace Settings).
Select your target provider (e.g. Anthropic or Google Gemini).
Enter your secret API key.
Toggle visibility with the eye icon or use the 1-click copy button to verify credentials.
Save settings. All subsequent AI requests from users in your workspace will route through your dedicated enterprise quota.